Switch KeyValSpine to flatcontainer

Signed-off-by: Moritz Hoffmann <[email protected]>

Author: antiguru

Cargo.lock

@@ -176,15 +176,22 @@ debug = 2
 # tend to get rewritten or disappear (e.g., because a PR is force pushed or gets
 # merged), after which point it becomes impossible to build that historical
 # version of Materialize.
+[patch."https://github.com/TimelyDataflow/timely-dataflow"]
+# Projects that do not reliably release to crates.io.
+timely = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
+timely_bytes = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
+timely_communication = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
+timely_container = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
+timely_logging = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
 [patch.crates-io]
 # Projects that do not reliably release to crates.io.
 timely = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
 timely_bytes = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
 timely_communication = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
 timely_container = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
 timely_logging = { git = "https://github.com/MaterializeInc/timely-dataflow.git" }
-differential-dataflow = { git = "https://github.com/MaterializeInc/differential-dataflow.git" }
-dogsdogsdogs = { git = "https://github.com/MaterializeInc/differential-dataflow.git" }
+differential-dataflow = { git = "https://github.com/antiguru/differential-dataflow.git", branch = "consolidate_layout_merger_chunk" }
+dogsdogsdogs = { git = "https://github.com/antiguru/differential-dataflow.git", branch = "consolidate_layout_merger_chunk" }
 
 # Waiting on https://github.com/sfackler/rust-postgres/pull/752.
 postgres = { git = "https://github.com/MaterializeInc/rust-postgres" }

Cargo.toml

@@ -1,4 +1,3 @@
-
 # cargo-vet audits file
 
 [criteria.maintained-and-necessary]
@@ -281,7 +280,7 @@ version = "23.5.26"
 [[audits.flatcontainer]]
 who = "Moritz Hoffmann <[email protected]>"
 criteria = "safe-to-deploy"
-version = "0.4.1"
+version = "0.5.0"
 
 [[audits.fluent-uri]]
 who = "Nikhil Benesch <[email protected]>"

misc/cargo-vet/audits.toml

@@ -36,6 +36,7 @@
 use std::any::Any;
 use std::cmp::Ordering;
 use std::fmt::Display;
+use std::sync::Arc;
 use std::time::Duration;
 
 use anyhow::Context;
@@ -109,7 +110,7 @@ where
         }
     }
 
-    match initialize_networking_from_sockets(sockets, process, workers, Box::new(|_| None)) {
+    match initialize_networking_from_sockets(sockets, process, workers, Arc::new(|_| None)) {
         Ok((stuff, guard)) => {
             info!(process = process, "successfully initialized network");
             Ok((

src/cluster/src/communication.rs

@@ -30,7 +30,7 @@ mz-compute-types = { path = "../compute-types" }
 mz-dyncfg = { path = "../dyncfg" }
 mz-dyncfgs = { path = "../dyncfgs" }
 mz-expr = { path = "../expr" }
-mz-ore = { path = "../ore", features = ["async", "flatcontainer", "process", "tracing_"] }
+mz-ore = { path = "../ore", features = ["async", "differential", "flatcontainer", "process", "tracing_"] }
 mz-persist-client = { path = "../persist-client" }
 mz-persist-types = { path = "../persist-types" }
 mz-repr = { path = "../repr" }

src/compute/Cargo.toml

@@ -23,7 +23,7 @@ use timely::progress::Timestamp;
 use timely::Container;
 
 use crate::logging::compute::ComputeEvent;
-use crate::typedefs::{KeyAgent, KeyValAgent, RowAgent, RowRowAgent, RowValAgent};
+use crate::typedefs::{KeyAgent, RowAgent, RowRowAgent, RowValAgent};
 
 /// Extension trait to arrange data.
 pub trait MzArrange: MzArrangeCore
@@ -270,36 +270,6 @@ where
     }
 }
 
-impl<G, K, V, T, R> ArrangementSize for Arranged<G, KeyValAgent<K, V, T, R>>
-where
-    G: Scope<Timestamp = T>,
-    G::Timestamp: Lattice + Ord + Columnation,
-    K: Data + Columnation,
-    V: Data + Columnation,
-    T: Lattice + Timestamp,
-    R: Semigroup + Ord + Columnation + 'static,
-{
-    fn log_arrangement_size(self) -> Self {
-        log_arrangement_size_inner(self, |trace| {
-            let (mut size, mut capacity, mut allocations) = (0, 0, 0);
-            let mut callback = |siz, cap| {
-                size += siz;
-                capacity += cap;
-                allocations += usize::from(cap > 0);
-            };
-            trace.map_batches(|batch| {
-                batch.storage.keys.heap_size(&mut callback);
-                batch.storage.keys_offs.heap_size(&mut callback);
-                batch.storage.vals.heap_size(&mut callback);
-                batch.storage.vals_offs.heap_size(&mut callback);
-                batch.storage.times.heap_size(&mut callback);
-                batch.storage.diffs.heap_size(&mut callback);
-            });
-            (size, capacity, allocations)
-        })
-    }
-}
-
 impl<G, K, T, R> ArrangementSize for Arranged<G, KeyAgent<K, T, R>>
 where
     G: Scope<Timestamp = T>,
@@ -415,8 +385,8 @@ mod flatcontainer {
     use differential_dataflow::lattice::Lattice;
     use differential_dataflow::operators::arrange::Arranged;
     use differential_dataflow::trace::TraceReader;
-    use mz_ore::flatcontainer::MzRegionPreference;
-    use timely::container::flatcontainer::{IntoOwned, Push, Region, ReserveItems};
+    use mz_ore::flatcontainer::{MzRegion, MzRegionPreference};
+    use timely::container::flatcontainer::{IntoOwned, Region};
     use timely::dataflow::Scope;
     use timely::progress::Timestamp;
     use timely::PartialOrder;
@@ -429,31 +399,10 @@ mod flatcontainer {
         Self: Clone,
         G: Scope<Timestamp = T::Owned>,
         G::Timestamp: Lattice + Ord + MzRegionPreference,
-        K: Region
-            + Clone
-            + Push<<K as Region>::Owned>
-            + for<'a> Push<<K as Region>::ReadItem<'a>>
-            + for<'a> ReserveItems<<K as Region>::ReadItem<'a>>
-            + 'static,
-        V: Region
-            + Clone
-            + Push<<V as Region>::Owned>
-            + for<'a> Push<<V as Region>::ReadItem<'a>>
-            + for<'a> ReserveItems<<V as Region>::ReadItem<'a>>
-            + 'static,
-        T: Region
-            + Clone
-            + Push<<T as Region>::Owned>
-            + for<'a> Push<<T as Region>::ReadItem<'a>>
-            + for<'a> ReserveItems<<T as Region>::ReadItem<'a>>
-            + 'static,
-        R: Region
-            + Clone
-            + Push<<R as Region>::Owned>
-            + for<'a> Push<&'a <R as Region>::Owned>
-            + for<'a> Push<<R as Region>::ReadItem<'a>>
-            + for<'a> ReserveItems<<R as Region>::ReadItem<'a>>
-            + 'static,
+        K: MzRegion,
+        V: MzRegion,
+        T: MzRegion,
+        R: MzRegion,
         K::Owned: Clone + Ord,
         V::Owned: Clone + Ord,
         T::Owned: Lattice + for<'a> PartialOrder<<T as Region>::ReadItem<'a>> + Timestamp,

src/compute/src/extensions/arrange.rs

@@ -130,7 +130,7 @@ pub(super) fn construct<A: Allocate>(
         let stream_to_collection = |input: Stream<_, ((usize, ()), Timestamp, Diff)>, log, name| {
             let mut packer = PermutedRowPacker::new(log);
             input
-                .mz_arrange_core::<_, KeyValSpine<_, _, _, _>>(
+                .mz_arrange_core::<_, KeyValSpine<usize, (), Timestamp, Diff, _>>(
                     Pipeline,
                     &format!("PreArrange Differential {name}"),
                 )

src/compute/src/logging/differential.rs

@@ -16,6 +16,7 @@ use mz_compute_client::logging::{LogVariant, LoggingConfig};
 use mz_ore::flatcontainer::{MzRegionPreference, OwnedRegionOpinion};
 use mz_repr::{Diff, Timestamp};
 use mz_storage_types::errors::DataflowError;
+use mz_timely_util::containers::PreallocatingCapacityContainerBuilder;
 use mz_timely_util::operator::CollectionExt;
 use timely::communication::Allocate;
 use timely::container::flatcontainer::FlatStack;
@@ -184,10 +185,8 @@ impl<A: Allocate + 'static> LoggingContext<'_, A> {
 
     fn reachability_logger(&self) -> Logger<TrackerEvent> {
         let event_queue = self.r_event_queue.clone();
-        let mut logger = BatchLogger::<
-            CapacityContainerBuilder<FlatStack<ReachabilityEventRegion>>,
-            _,
-        >::new(event_queue.link, self.interval_ms);
+        type CB = PreallocatingCapacityContainerBuilder<FlatStack<ReachabilityEventRegion>>;
+        let mut logger = BatchLogger::<CB, _>::new(event_queue.link, self.interval_ms);
         Logger::new(
             self.now,
             self.start_offset,

src/compute/src/logging/initialize.rs

@@ -19,10 +19,10 @@ use mz_ore::cast::CastFrom;
 use mz_ore::flatcontainer::{MzRegionPreference, OwnedRegionOpinion};
 use mz_ore::iter::IteratorExt;
 use mz_repr::{Datum, Diff, RowArena, SharedRow, Timestamp};
+use mz_timely_util::containers::PreallocatingCapacityContainerBuilder;
 use mz_timely_util::replay::MzReplay;
 use timely::communication::Allocate;
 use timely::container::flatcontainer::FlatStack;
-use timely::container::CapacityContainerBuilder;
 use timely::dataflow::channels::pact::Pipeline;
 
 use crate::extensions::arrange::{MzArrange, MzArrangeCore};
@@ -57,7 +57,7 @@ pub(super) fn construct<A: Allocate>(
         );
         type UpdatesRegion = <((UpdatesKey, ()), Timestamp, Diff) as MzRegionPreference>::Region;
 
-        type CB = CapacityContainerBuilder<FlatStack<UpdatesRegion>>;
+        type CB = PreallocatingCapacityContainerBuilder<FlatStack<UpdatesRegion>>;
         let (updates, token) = Some(event_queue.link).mz_replay::<_, CB, _>(
             scope,
             "reachability logs",
@@ -102,7 +102,7 @@ pub(super) fn construct<A: Allocate>(
                 );
 
                 let updates =
-                    updates.as_collection(move |(update_type, addr, source, port, ts), _| {
+                    updates.as_collection(move |(&update_type, addr, &source, &port, ts), _| {
                         let row_arena = RowArena::default();
                         let update_type = if update_type { "source" } else { "target" };
                         let binding = SharedRow::get();
@@ -118,7 +118,7 @@ pub(super) fn construct<A: Allocate>(
                             Datum::UInt64(u64::cast_from(port)),
                             Datum::UInt64(u64::cast_from(worker_index)),
                             Datum::String(update_type),
-                            Datum::from(ts.clone()),
+                            Datum::from(ts.copied()),
                         ];
                         row_builder.packer().extend(key.iter().map(|k| datums[*k]));
                         let key_row = row_builder.clone();